Data connections let you connect the variables in your model to data sources. In most cases, this is used to pull in historical data.
When a variable is linked to data, it will take its value from the data if such a value exists, and will fall back to its own formula if a value doesn't exist. You can use this to perform forecasts based on historicals.
In the example below, we have data up until May'22 (shown by the data icon in each month), and from Jun'22 onwards, Causal falls back to the formula.